VENDORiQ: Google Deftly Side-Steps the Data Mesh Bandwagon. You Should Too.
15 March 2022: Google announced general availability of Dataplex, a ‘dash fabric’ (aka, data mesh) solution that allows enterprises to centrally manage and control data across data lakes, databases and data warehouses. Google claims that the product can mesh Google Cloud with open source technology for analytics professionals to better govern data. Dataplex was launched together with Datastream and Analytics Hub that make up Google Cloud’s database services in its analytics portfolio.
Why it’s Important
Since it was introduced in 2019 by its creator Zhamak Dehghani, the concept of data mesh is becoming hyped. Similar to service oriented architecture (SOA), it is misunderstood by vendors and buyers alike who believe that it is all about technology. Instead, data mesh is as much a philosophical shift, from viewing data being centralised in data lakes or warehouses, to managing data close to where it is created, through a domain-oriented design with a self-serve data infrastructure.
Google, on the other hand, identifies Dataplex as a data fabric, which provides a technology layer over disparate data sources for better access, discovery, integration, governance and security. Data fabric focuses on existing multiple centralised technologies that consolidate data. Data mesh, on the other hand, promises a fully domain-oriented, decentralised approach, since it considers all enterprise data as a set of different repositories, preventing any loss in domain expertise during translation, unlike with a data lake. Thus, in a pure data mesh platform, different groups can manage data as they see fit, sharing some common governance measures while maintaining their domain knowledge on the data.
IBRS has observed that data catalogue vendors are leveraging data mesh rhetoric to market their products. However, most of these do not truly align with the philosophy of data mesh, which is fine for the near term, as few organisations are prepared for the changes involved when adopting such a democratised approach to data management.
- Analytics teams
Organisations that want to explore data mesh concepts must carefully consider shifts in data team structures, roles, responsibilities and skills before looking at technical solutions.Some changes brought by such a data architecture approach will impact domain-specific variations in data across departments, domain ownership, data product self-containment, and governance architecture to preserve global controls.
Related IBRS Advisory
1. Business First Data Analytics - Webinar and Q&A